Diagnosing and treating heart rhythm disorders often involve the introduction of a catheter having a plurality of sensors/probes into a cardiac chamber through the surrounding vasculature. The sensors detect electric activity of the heart at sensor locations in the heart. The electric activity is generally processed into electrogram signals that represent signal propagation through cardiac tissue at the sensor locations.
Systems can be configured to display the electrical signals detected in the cardiac chamber as an activation map based on voltages detected. Robust and reliable estimation of onset times for sensed activation signals is the key to visualizing underlying activation patterns and identify targets for applying therapy, e.g. ablation therapy. Traditional methods use characteristic features of the activation signals such as steepest descent for a unipolar signal or most negative peak of a derivative unipolar signal. Any characteristic feature is susceptible to ambiguity stemming from noise or other artifacts, such as a far-field activation signal or multiple large negative peaks adjacent to one another, which superimpose over the signal of interest. There exists a need to improve the reliability of detection of onset times of activation signals in anatomical mapping.